• No results found

Success, survive or escape? Aspirations and poverty traps.

N/A
N/A
Protected

Academic year: 2021

Share "Success, survive or escape? Aspirations and poverty traps."

Copied!
36
0
0

Loading.... (view fulltext now)

Full text

(1)

Durham Research Online

Deposited in DRO:

19 September 2017

Version of attached le:

Accepted Version

Peer-review status of attached le:

Peer-reviewed

Citation for published item:

Chivers, D. (2017) 'Success, survive or escape? Aspirations and poverty traps.', Journal of economic behavior and organization., 143 . pp. 116-132.

Further information on publisher's website:

https://doi.org/10.1016/j.jebo.2017.09.018 Publisher's copyright statement:

c

2017 This manuscript version is made available under the CC-BY-NC-ND 4.0 license http://creativecommons.org/licenses/by-nc-nd/4.0/

Use policy

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, for personal research or study, educational, or not-for-prot purposes provided that:

• a full bibliographic reference is made to the original source • alinkis made to the metadata record in DRO

• the full-text is not changed in any way

The full-text must not be sold in any format or medium without the formal permission of the copyright holders. Please consult thefull DRO policyfor further details.

(2)

Success, Survive or Escape? Aspirations and

Poverty Traps

David Chivers

Abstract

I present a model of occupational choice where an agent decides whether to invest in a project that yields risky returns or a project that yields safe returns. An agent’s utility is a¤ected by the presence of an aspiration level which will only be satis…ed if their …nal income is above the poverty line. I show that agents who are su¢ ciently above the poverty line will invest in the risky project and are able to aspire for success. An agent, however, who is just above the poverty line, may be so concerned about falling into poverty that they choose to invest in the safe project. These individuals aspire only to survive. Alternatively, if an agent is su¢ ciently below the poverty line, then they will invest in the risky project even if expected returns are lower than the safe project. These individuals have "nothing left to lose" and therefore aspire to escape. Two forms of poverty traps emerge from the resulting equilibria: one above the poverty line, and one below the poverty line. Finally, I o¤er empirical support for the model based on individual level survey data across a large number of countries.

JEL Classi…cation: D31, D81, E24, L26, O11.

Keywords: Poverty Traps, Entrepreneurship, Aspirations, Loss aversion, Development.

I would like to thank Keith Blackburn, Matthew McKernan, Debraj Ray and Les Reinhorn as well as the anonomous referees for their insightful and helpful comments. I am grateful to the Royal Economic Society for providing funding to present this work at the Development Economics and Policy conference 2017 at the University of Göttingen. All errors remain the responsibility of the author. Address for correspondence: David Chivers, Department of Economics and Finance, Durham University Business School, Durham University, Mill Hill Ln, Durham DH1 3LB, England. Tel: 0191 334 5200 E-mail: david.chivers@durham.ac.uk.

(3)

1

Introduction

What does it mean to have aspirations? We tend to think of an aspiration as above and beyond the position in life we are in now: to achieve great things, to become rich, to be a success. However, for some individuals, the aim is simply to survive, to maintain the status quo. This is born out of a fear that falling below their status quo is far worse than the current state they are in. For others, the fear of the status quo is already being felt, and will take any chance to escape their current situation. What aspirations have in common, however, is that they are formed in the present about the kind of future we want and the kind of future we hope will never happen. As a consequence, an aspiration is de…ned by the relative weights we attach to the overall probability of success and failure for a desired objective.

The natural question one turns to is how we form our aspirations. One could argue that our aspirations are heterogenous and innate, shaped by our parents, or perhaps our culture. Despite our own, idiosyncratic aspirations, we may also share a common aspiration, to be free from poverty, for example. If this is the case, the e¤ect of this common aspiration will di¤er depending on our own proximity to the poverty line. This will have important implications for wealth-enhancing investment decisions.

For example, starting a business can make individuals extremely rich or leave them desperately poor. Faced with the decision either to start a business or obtain a safer form of income, someone who is relatively wealthy can aspire for success, and may see this as a risk worth taking. Someone who is just above the poverty line, however, may view this business opportunity as a risk too far. This desire to maintain the status quo is born out of a fear that if their business failed they would fall into poverty. Someone who is already below the poverty line will have "nothing left to lose" and may view starting a business as the only way to escape poverty. This paper will show how individual aspirations to avoid poverty can result in poverty traps.

Within the theoretical poverty trap literature, many of the results are driven by moves away from the standard neoclassical paradigm, in order to create multiple equilibria and path-dependence (i.e., long-run outcomes are dependent on initial conditions). These include non-convexities in technolo-gies, market incompleteness (often in the form of credit constraints) and imperfect functioning of institutions (see Azariadis 1996; 2005 for surveys).

More recently there has been a focus on the role of aspirations in the creation and persistence of poverty traps (see Ray, 2006 for an introduction). Aspirations may be conditioned by relative economic status, as in Moav and Neeman (2010) and Ray and Robson (2012), or by parents’aspirations for their children’s education, as in Mookherjee et al. (2010). For example,

(4)

Genicot and Ray (2017) develop a model in which there is an endogenous re-lationship between economic outcomes and individual aspirations, and hence income and the distribution of income are jointly determined.

Dalton et al. (2016) show how a poverty trap can occur when e¤ort a¤ects …nal wealth. The optimal amount of e¤ort chosen will be based on whether an individual believes they can meet their aspiration level. As a result, aspiration failure will be self-fulling. Blackburn and Chivers (2015) show how the e¤ects of aspirationally-induced loss aversion can result in persistent inequality due to the fear or falling below a certain level of income. However, all of these models treat aspirations as unidirectional: either success (as in Genicot and Ray, 2017 and Dalton et al., 2016) or survival (as in Blackburn and Chivers, 2015).

The key insight of this paper is that aspirations should be thought of as multidirectional. That is to say, the e¤ect of aspirations will di¤er depending on the situation an individual faces. The model developed within this paper will show that an individual’s aspiration to avoid poverty will result in di¤er-ent behaviour depending on how far away the individual is from experiencing poverty, or how close the individual is from escaping poverty. The result of this model is that two poverty traps emerge: one below the poverty line and one just above the poverty line.

The notion that aspirations of the poor may di¤er, can be traced back to Banerjee (2000), who suggests that there are two distinct and competing views of poverty: "poverty as desperation" and "poverty as vulnerability". Banerjee (2000) argues that, if we view poverty as a form of desperation, the poor would wish to invest in wealth-enhancing projects, but may be denied credit. However, if one views poverty as vulnerable individuals facing the possibility of falling further into poverty, then the poor may forgo investment in a risky project. This paper di¤ers to Banerjee’s (2000), as it shows how both these types of behaviours can stem from aspirations alone, without the need of appealing to non-convexities in technologies.

The approach of this paper is driven by recent advancements in decision theory - speci…cally, aspiration levels. Aspiration levels occur when individ-uals are faced with a risky prospect. The individual evaluates the project based on their weighted preferences of the overall probability of success or failure. What individuals deem to be a success or failure is judged with respect to some aspirational outcome (see Diecidue and Van de Ven, 2008).

Aspiration levels are similar to reference points that occur in loss-averse preferences developed by Kahneman and Tversky (1979) in their seminal work on prospect theory. Loss aversion is the notion that individuals have a stronger preference to avoid losses than to obtain gains, relative to a particu-lar reference point. Although there is a subtle di¤erence between aspiration

(5)

levels and reference points, the two are linked (see Lopes and Oden, 1999, for a comparison). Aspiration levels are based on probabilities and view out-comes as …nal states. Conversely, reference points are a behavioural concept linked to changes in wealth. These features give rise to a discontinuity in the utility function for aspiration levels and a kinked utility function under loss aversion.

The presence of aspirational levels in risky projects has been found in a number of experimental and empirical studies (see e.g., Holthausen 1981; Mezias 1988; Langer and Weber 2001; Mezias et al. 2002). There is also a growing literature examining aspirations in developing countries. Pasquier-Doumer and Brandon (2015) examine educational investment among nous and non-indigenous children in Peru. They …nd that although indige-nous children have lower aspirations than non-indigeindige-nous children, this is mostly explained by the e¤ects of socioeconomic status. Aspirations have also been the subject of a number of randomised control trials (see Bogliancino and Gozález-Gallo, 2015, Macours and Vakis, 2009). Bernard and Seyoum Ta¤esse (2014) conducted an experiment in order to examine peer e¤ects in the formation of aspirations in rural Ethiopia. The treatment group of indi-viduals were invited to watch a documentary about successful entrepreneurs from similar communities. A …rst control group watched an entertainment programme, while a second control group were simply surveyed. They found that aspirations were higher among the treated, but unchanged amongst the control groups.

The purpose of this paper is to highlight the importance of aspirations for the development process, as risky, wealth-enhancing investment decisions are key to understanding the creation and persistence of poverty traps.

The remainder of the paper is outlined as follows. In section 2, I examine the e¤ect of aspirations on wealth-enhancing investment decisions in a sto-chastic overlapping generation model. In section 3, I test the implications of this model by using individual-level survey data across a large sample of countries and create a linear probability model in order to identify a common aspiration level among entrepreneurs. Finally, I o¤er a conclusion in section 4.

2

The Model

Consider a small, open economy with a unit population of two-period lived agents in an overlapping generations framework. Generations are connected through transfers of a productive technology such as human capital (as in Lucas, 1988), or stock of knowledge (as in Romer, 1986). In the …rst

(6)

pe-riod of life, young agents choose to invest in a risky project or invest in a safe project with guaranteed returns. The risky project may be opportunity-driven (which yields higher returns on average than the safe project) or ne-cessity driven (which yields lower returns on average than the safe option). In the …nal period of life, old agents consume the realised output based on their investment choices. Agents utility is governed by a common aspiration level in the form of Diecidue and Van de Ven (2008).

2.1

Preferences

Agents have identical preferences and derive utility from lifetime income such that ut= u(xt+1): Following Diecidue and Van de Ven (2008), agents have a common aspiration level in their utility function x such that their expected utility is given as Vt= E(xt+1 x ) + P (xt+1 x ) P (xt+1 < x ): The term P (xt+1< x ); is the probability of failing to attain their aspiration, and P (xt+1 x ); is the probability of achieving their aspiration. Whether one views an aspiration as a disutility of failure of being below x ; or an added utility of success from being above x ; is inconsequential to the model. The only requirement is that the combined weights of success and failure do not cancel each other out (where for = ): Consequently, I set = 0 in order to simplify the model, and therefore assume agents face an added disutility based on the probability of failure to meet their aspiration level of income.1

Let xt+1 and x denote, the actual and aspiration levels of income of an agent, respectively. The expected payo¤ to each agent is given by

Vt= E(xt+1 x ) P (xt+1< x ); (1)

where > 0.

2.2

Technologies

There exists two types of risky projects that require a …xed amount of capital k > 0 on start-up: an "opportunity-driven" project, where it = kB, and a "necessity driven" project, where it = kb. If agents do not invest in a risky project, they can invest in a safe project which requires no capital investment, where it = 0:As agents are not endowed with any physical capital at the start of life, agents must borrow to start a business (as in Banerjee and Newman,

1It is trivial to show what happens when > 0 : Adding this to the model simply

(7)

1993; Galor and Zeira, 1993). Credit markets, however, function perfectly, and hence capital is borrowed at the world rate of interest r.2

The formation of the next period’s productive technology ht+1will depend on a fraction of last period’s productive technology ht; where 2 (0; 1). An opportunity-driven entrepreneur’s productive technology will increase by B from investment, whereas a necessity-driven entrepreneur will increase by b: An individual who does not start a business and pursues the safe project will increase their productive technology by s: The hierarchy of investments is given as B > s > b. The evolution of the productive technology will diverge depending on the investment decision of the agent such that

ht+1= 8 < : ht+ B if it= kB; ht+ s if it = 0; ht+ b if it= kb: (2) Although next period’s productive technology ht+1 is known, the next period’s output yt+1 is uncertain for risky projects. This will be important for the agent’s investment decision as output yt+1is simply the agents income xt+1: Output for both risky projects and the safe project is expressed as

yt+1=

A(1 + t+1)ht+1 if it = kB or it= kb;

Aht+1 if it= 0; (3)

where total factor productivity A is a positive constant. The term t+1 rep-resents a technology shock and is designed to re‡ect the fact that the returns to entrepreneurship are comparatively risky compared to paid employment.3 The shock t+1; is uniformly distributed on the interval ( a; a); with probability density function f ( t+1) = 2a1(a < 1):4 This implies that the expected value of this shock is zero E[ t+1] =

a Z

a

t+1f ( t+1)d t+1 = 0; and

that the variance (and therefore risk) of the shock, is determined by a which is

2Blackburn and Chivers (2015) show how credit market frictions and

aspirationally-induced loss aversion are isomorphic. Hence, the purpose of assuming perfectly functioning credit markets is so that one can see clearly the e¤ect of aspirations on inequality. For an example of a model in which entrepreneurs face a distorted credit market, see Antunes et al. (2015).

3It is possible to rewrite this model with risk entering the productive technology rather

than output. The subsequent analysis would be unchanged with the only di¤erence being the transitional dynamics of the model.

4One could argue that the cost of the project, and level of risk, may vary between an

opportunity-driven project and a necessity-driven project. However, I assume the cost of the project, and the level of risk, is the same for simplicity of exposition.

(8)

given as var ( t+1) = E ( t+1 E ( t+1))2 = E( 2t+1) = a Z a 2 t+1f ( t+1)d t+1= a2 3 :

2.3

Incomes

Using equations (2) and (3); and noting an old agent’s output yt+1 is their income xt+1, an agent can obtain the following:

xt+1 = 8 < : A(1 + t+1)( ht+ B) (1 + r)k if it = kB; A( ht+ s )if it= 0; A(1 + t+1)( ht+ b) (1 + r)k if it= kb: (4) Correspondingly, the expected …nal income becomes

E [xt+1] = 8 < : A( ht+ B) (1 + r)k if it = kB; A( ht+ s) if it = 0; A( ht+ b) (1 + r)k if it= kb. (5) The evolution of incomes captures a number of stylised facts about the entrepreneurial earnings distribution.5 Although the returns to entrepre-neurship are lower than paid employment on average, the returns to entre-preneurship have a much wider distribution (see Lin et al., 2000, Hamilton, 2000, and Herranz et al. 2015). Furthermore, De Nardi et al. (2007) show that if we discount the solely self-employed from entrepreneurial earnings, average entrepreneurial earnings are higher than wage earnings. Finally, en-trepreneurs tend to be drawn disproportionately from each end of the ability distribution and are remunerated accordingly (Åstebro et al., 2011). Data on entrepreneurial earnings are reported in Table A1.

I assume that on average the opportunity-driven, risky project yields a greater return that the safe project such that B > s: To make the analysis non-trivial, however, I assume that the presence of uncertainty (via t+1) in the production of the opportunity-driven project means that it is not for certain that the risky project will always guarantee a higher level of income than the safe project. I also assume that the necessity-driven project will on average be lower than the safe project such that s > b, but there is a possi-bility once production takes place that the necessity-driven project can yield

5With some further parameter restrictions one could trivially use a convex function that

represents all entrepreneurial activities - returns start o¤ smaller than the safe project and eventually eclipse them. However, for demonstrative purposes, I use the above with no loss of generality.

(9)

a higher income. As a result of these two assumptions, the parameter restric-tions B(1 a) sand b(1+a) sare needed. If these parameter restrictions were not in place, then the initial level of productive technology would play no role in the individual’s investment decision and the decision to invest would no longer be risky. Speci…cally, the opportunity-driven project would always yield more than the safe project, and the necessity-driven project would always yield less than the safe project. To simplify the problem, and to ensure risk plays a role in the investment decision, I assume that agents cannot create insurance markets in order to mitigate these risks.

I now analyse the occupational choice facing individuals above and below the poverty line.

2.4

Non-poor Agents

A non-poor agent is de…ned as an individual who can be above the poverty line x if they invest in the safe project, such that A( ht+ s ) x : The choice this agent faces will be whether to invest in the opportunity-driven project or the safe project. This is because both of these investments will always be preferable to the necessity-driven project, as the expected value of opportunity-driven project is always higher than a necessity-driven project, B > b. As individuals consume all output in the last period of life, one can simply substitute equation (5) into the utility function in equation (1). For agents that invest in the safe project their expected utility becomes

Vtjit=0 = A[ ht+ s] x : (6)

Notice that as non-poor agents are already above their aspiration level, in-vesting in a safe project does not result in any disutility associated with failure.

By contrast, suppose that the decision to invest in the opportunity-driven project may, or may not, satisfy an individual’s aspiration level (i.e., A(1 + t+1)( ht+ B) (1 + r)k x ), or A(1 + t+1)( ht+ B) (1 + r)k < x ): It is now possible to de…ne a critical value of the productivity shock bt+1 such that an individual would meet their aspiration, if t+1 bt+1;or fail to meet their aspiration if, t+1< bt+1:In order to determinebt+1 one must …nd the the level of t+1 which satis…es the following equation

A(1 +bt+1)( ht+ B) (1 + r)kB = x : (7)

Accordingly, the probability of failing to attain aspirations is P (xt+1 < x ) = P (xt+1 < bt+1) =

Rbt+1

a f ( t+1)d t+1 = bt+1+

(10)

into equation (1), the expected utility of an individual that invests in the risky, opportunity-driven project is given as

Vtjit=kB = A ( ht+ B) (1 + r)k

bt+1+ a

2a : (8)

The decision to invest in the opportunity-driven project rests on whether the opportunity-driven project yields higher utility than the safe project, where Vtjit=k > Vtjit=0: By using equations (6) and (8), this condition is

given as

A (B s) (1 + r)k bt+1+ a

2a : (9)

Rearranging the expression in equation (7) delivers bt+1 =

(1 + r)k + x A( ht+ B) A( ht+ B)

(ht); (10)

where h(ht) < 0. Given the above, the condition for investment in the opportunity-driven project is expressed as A(B s) (1 + r)k (ht)+a

2a :

This condition is more likely to be satis…ed the higher an agent’s initial knowledge ht, the lower the weight of aspirations ; and the lower the level of uncertainty a.6

The condition which determines whether or not a non-poor agent invests in an opportunity-driven project is de…ned for any ht; such that ht bhB; in

A (B s) (1 + r)k 0 @ bhB + a 2a 1 A : (11)

2.5

Poor Agents

A poor agent is de…ned as an individual whose income cannot be above the poverty line x if they invest in the safe project, such that A( ht+ s ) < x :I assume, initially, that poor individuals do not have access to the opportunity-driven project. This is not a crucial assumption for the model but is imple-mented to highlight how aspirations will a¤ect poor and non-poor agents

6The e¤ect of a follows b

t+1 < 0 in (8). This is due to the fact that for a non-poor

agent A( ht+ B) (1 + r)k x > 0 must hold. If this condition does not hold, then

investment in the opportunity-driven project will never be chosen, as the expected income from this investment would be less than the aspiration level an agent. Or in other words, they would always choose to invest in the safe asset.

(11)

di¤erently. As mentioned earlier, one could replace both risky projects with a convex function of knowledge accumulation: returns to entrepreneurship start o¤ smaller than the safe project and eventually outgrow them. Imposing the assumption that poor individuals do not have access to the opportunity-driven project is equivalent to imposing further parameter restrictions on a model with a convex function of knowledge accumulation. Nevertheless, I return to the consequences of relaxing this assumption in the discussion of equilibrium outcomes.

If a poor agent invests in the safe project, then the presence of the disutil-ity of being below their aspiration level will always be felt. By using equations (5) and (1), the poor agent’s expected utility when choosing the safe project becomes

Vtjit=0 = A[ ht+ s] x : (12)

Although on average the necessity-driven project yields a lower return than the safe project, as s > b; this is not a certainty. If a poor agent chooses the necessity-driven project then there is a chance that the project may take them over the poverty line x (i.e., A(1 + t+1)( ht+ b) (1 + r)k x ), or A(1 + t+1)( ht+ b) (1 + r)k < x ): Hence, it is possible to …nd a critical value of the productivity shock bt+1 that takes the individual’s …nal income over the poverty line. The expected utility of the necessity-driven project is therefore given as

Vtjit=kb = A ( ht+ b) x (1 + r)k

bt+1+ a

2a : (13)

The occupational choice for the poor agent is determined by comparing the respective utilities of the necessity-driven project and the safe project. Substituting (13) and (12) in Vtjit=kb Vtjit=0; one obtains the expression

that would result in the agent having a higher utility under necessity-driven entrepreneurship than the safe project

a bt+1

2a A[s b] + (1 + r)k: (14)

I therefore determine bt+1 in a similar manner to non-poor agents in (7) noting h(ht) < 0. The condition a 2a(h) A[s b] + (1 + r)k is more likely to hold the higher the poor agent’s initial level of knowledge ht. As poor agents move closer towards their aspirational level, they are more likely to invest in a project that would see them escape poverty. As a result, this condition is more likely to be satis…ed the greater the strength of aspirations

(12)

If the poor agent chooses the safe project or fails to meet their aspira-tion under the necessity-driven project, they will still experience the negative utility of being below their aspiration. Accordingly the condition for invest-ment the necessity-driven for a poor agent is de…ned for any ht; such that

ht bhb; in 0 @a bhb 2a 1 A A[s b] + (1 + r)k: (15)

2.6

Equilibrium Outcomes

There exists a critical value htwhich determines whether an individual can, or cannot, be above the poverty line x : This critical value determines whether the agent is non-poor (A( ht+ s ) x ) ; or poor (A( ht+ s ) < x ) ; and is expressed as

bhx =

x sA

A : (16)

Using this result together with equations (11) and (15); it is possible to de…ne the threshold levels of the initial productive technology bhb; bhx and bhB. These threshold levels of the initial productive technology determine the long-run consequence of each generation’s investment decisions. The evolution of this process for all initial levels of knowledge is described by

ht+1 = 8 > > > < > > > : ht+ s if ht< bhb; ht+ b if bhb ht < bhx ; ht+ s if bhx ht < bhB; ht+ B if ht bhB: (17)

Depending on the initial level of knowledge of the …rst generation ht; the economy will converge to high, medium or low steady-state (hH = 1B , hM = 1s or hL = 1b ):

7 The dynamics of this process are shown in Figure 1.

[Figure 1 about here]

7One could view this model as either understanding the emergence of inequality across

economies (as this analysis does), or the long-run persistence of inequality within an economy. The transitional dynamics depend on the initial level of knowledge for the former and on the initial distribution of knowledge within an economy for the latter.

(13)

If an individual has an initial productive technology ht < bhb; then they and their future generations will invest in the safe project. This occurs until the accumulation of knowledge over generations reaches bhb where that particular generation will invest in a necessity-driven project. Even though on average this project will yield lower returns than the safe project, due to the presence of risk in the production function it is possible that some generations will reach their aspiration level and escape poverty. However, as agents only pass on knowledge to their future dynasties, any favourable productivity shock that allows a generation to escape poverty will only bene…t that generation. In this model, knowledge determines future generation’s productivity, not …nal income. Hence, dynasties whose initial productive technology is ht < bhx ; will converge to a low steady-state hL; and always invest in necessity-driven projects.

Individuals with initial productive technology ht bhx have the choice of investing in an opportunity-driven project that yields higher expected returns than the safe project. Dynasties whose initial productive technology is ht < bhB; are su¢ ciently concerned, however, by the possibility of falling below their aspiration level and hence invest in the safe project. It follows that dynasties whose initial productive technology is bhx ht < bhB will converge to the mid-level steady-state hM

Finally, if an individual has initial productive technology ht bhB; then they and their future generations will always invest in the opportunity-driven project. Consequently, these dynasties will converge to a high-level steady state hH: This is not to say that some generations will not fall into poverty during a bad productivity shock. The level of knowledge, however, they pass on, is not a¤ected by this bad productivity shock and so will not a¤ect the upward trajectory of their dynasties.

From equations (11) and (15) it can be shown that bhb (a; ) < 0 and bhB(a; ) > 0. Hence, increasing the level of risk (a) or the strength of aspirations ( ) moves bhb to the left and bhB to the right in Figure 1.

2.7

Discussion

One of the features of this model is that poverty traps can arise purely as a result of the nature of preferences, rather than the structure of technologies, or functioning of markets. This is not to say that credit market imperfections are unimportant as it is likely that both aspirations and liquidity constraints will ultimately have a role in the creation of poverty traps. In fact, adding a liquidity constraint to this model turns out to be isomorphic to the ad-dition of an aspiration level within the utility function (see Blackburn and

(14)

Chivers, 2015). This result highlights why it is di¢ cult to distinguish be-tween behavioural e¤ects and credit frictions empirically (see Carroll, 2001). Consequently, empirical results that do not take into account aspirational e¤ects may overestimate the role of credit frictions.

Another feature of this model is that the resulting equilibria are remi-niscent of Banerjee’s (2000) two types of poverty. Those with initial wealth su¢ ciently below the poverty line bhx are experiencing poverty of despera-tion. Those with initial wealth just above bhx are concerned about falling into poverty and hence will not invest in risky projects. This can be thought of as experiencing poverty of vulnerability.

The analysis of the model has so far concentrated on circumstances where aspiration levels enter an agent’s investment decision. If, however, aspirations are too high for the necessity-driven project, where the best outcome of the project would never satisfy the agent’s aspiration level, ht+ b(1 + a) (1 + r)k < x , individuals would simply always invest in the safe project. These agents will always su¤er the disutility of being below their aspiration level. Similarly, if aspirations were too low for the opportunity-driven project, then the worst outcome of the project would not take them below their aspiration level, ht+B(1 a) (1+r)k > x . These individuals would never experience the possibility of falling below their aspiration level and therefore always invest in the opportunity-driven project. Under these conditions, a policy that lowers aspirations that are too high may encourage individuals into a wealth-enhancing investment in order to avoid aspiration failure, such as in Dalton et al. (2016).

Finally, if the assumption that poor individuals do not have access to the opportunity-driven project is relaxed, then there will only be two steady states, hM and hH:This is because individuals below the poverty line bhx will always invest in the opportunity-driven project in order to escape poverty. Over generations, poor individuals will escape poverty and converge to the mid-level steady-state hM. One of the reasons for implementing this assump-tion is to show that su¢ ciently poor individuals would invest in risky invest-ments, even if the average returns of the investment are less than the safe project. In essence, they act as if they are risk seeking as they have "nothing left to lose". If this assumption is removed, however, one can also think of other reasons why poor individuals are less likely to invest in an opportunity-driven project, and more likely to invest in a necessity-opportunity-driven project. These reasons are explored in the next section.

(15)

3

Empirical Evidence

The model outlined above demonstrates how aspiration levels are important from a development perspective, as wealth-enhancing investments are often risky. The most obvious example of a wealth-enhancing investment that en-tails a substantive amount of risk is entrepreneurship. This is shown by the fact that the spread of income earnings is much larger for entrepreneurs than workers (see Hamilton, 2000; Lin et al., 2000). Furthermore, the probability of a business failing is much higher than being laid o¤ from a job.8

The importance of entrepreneurship as a source of innovation and growth dates back to Schumpeter (1911). A simple plot (shown in Figure 2) of the percentage of entrepreneurs within a country on a country’s level of development, however, shows a seemingly counter-intuitive relationship.9

[Figure 2 about here]

One of the reasons for this inverse relationship between entrepreneurship and development is due to problems in measuring what constitutes an entre-preneur. One can split the motivations for entrepreneurship into two types: opportunity-driven and necessity-driven (see Lunati, 2010). Opportunity-driven entrepreneurs are more likely to start …rms that are more pro…table, innovative and grow, compared to necessity-driven entrepreneurs. This is be-cause opportunity-driven entrepreneurs choose to take advantage of a busi-ness venture, whereas necessity-driven entrepreneurs, start a …rm because they have no other means of income (through paid employment, for exam-ple). These two types of entrepreneurs have di¤erent aspirations for their …rms. Opportunity-driven entrepreneurs aspire to improve their current sit-uation and can therefore aspire for success. Necessity-driven entrepreneurs feel they have no other option but to enter entrepreneurship and aspire to escape their current situation.

In order to understand what drives the aspirations of these two types of entrepreneurs, I take advantage of a unique data set, the Global Entrepre-neurship Monitor survey (GEM). The GEM survey has detailed information

8In the US for example, the Small Business Association (2016) report a 78.5% chance

of a …rm surviving past their …rst year, compared to a 1% probability of being laid o¤ (Bureau of Labor Statistics, 2016).

9Due to data availability, I use average rates of nascent entrepreneurs by country, across

the period 2001-2010, from GEM survey data, and use average GDP per capita from the World Bank. Wennekers et al. (2005) …nds a more intuitive U-shaped relationship between entrepreneurship and development, using a sample of 36 countries, within the GEM survey, in 2005. However, the upturn in entrepreneurial rates after a certain level of development is relatively small.

(16)

on start-up activities across a large number of individuals and spanning a wide variety of countries. The sample I use contains 44466 individual ob-servations, within 86 di¤erent countries, over a 10 year period (see Table A2 in the Appendix for the list of countries in the sample). One of the main advantages of the GEM survey is that it asks individuals their motivation in starting a business, and catches individuals at the start of the entrepreneur-ial process. I code individuals who report being in the process of starting a business due to "taking advantage of a business opportunity" as opportunity-driven, whereas those reporting starting a business due to a "lack of a better alternative", I code as necessity-driven. If I now plot the proportion of en-trepreneurs that report being opportunity-driven on development (Figure 3) we see a positive relationship.10

[Figure 3 about here]

One of the implications of the model, outlined in Section 2, is that those su¢ ciently above their aspiration level will take on an opportunity-driven project to aspire for success, and those individuals su¢ ciently below their aspiration level will take on a necessity-driven project and aspire to escape poverty. One way of testing this implication is considering the e¤ect of an individual’s income level on their motivation for starting a …rm. If an entrepreneur has a high income level, they would be less worried about falling into poverty, and therefore would be more likely to only take on opportunity-driven investments. If, however, an individual has a low in-come, they would arguably be more concerned about falling into poverty or escaping poverty. Therefore, low-income entrepreneurs are more likely to have a necessity driven motive. As a consequence, I test the hypothesis that the higher the entrepreneur’s income, controlling for other factors, the more likely they are to be opportunity-driven, and the less likely they are to be necessity-driven.

It is important to note that one of the di¢ culties in measuring aspirations is that they are not directly observable. As this analysis relies on self-reported data it may su¤er from a number of potential issues. For example, it is possi-ble that an entrepreneur may claim their motivation was opportunity driven (when it was in fact necessity driven) to save face in front of an interviewer. In spite of this, Bernard and Seyoum Ta¤esse (2014) argue that measuring aspirations by directly asking individuals may be preferable to an indirect approach. This because measuring aspiration indirectly can be problematic given the many factors which in‡uence an individual’s entrepreneurial deci-sion (as argued in Section 2.7).

10Simiarly, Ács and Varga (2005) …nd opportunity-driven entrepreneurship has a positive

(17)

3.1

Methodology

The GEM survey reports income entrepreneurs surveyed in three categories: low, medium and high, with respect to the distribution of the country in which they are surveyed. The income variable, therefore, captures aspirations to avoid relative poverty within a country, as opposed to relative poverty across countries or absolute poverty. To isolate the e¤ects of income on entrepreneurial motivation I control for a number of factors reported in Table 1.

[Table 1 about here]

I add dummy variables for the level of education obtained, as well as for age categories. This is because human capital and experience are likely to increase idea creation, which may lead to opportunity-driven entrepreneur-ship. Furthermore, those with high levels of human capital and experience may have better employment opportunities, which would reduce the chance of becoming a necessity entrepreneur. The employment status of individu-als will be of particular importance for the motivations of starting a …rm, especially whether an individual is unemployed. I also control for whether an individual answered "yes" or "no" to the statement "you have the knowl-edge, skill and experience required" to become an entrepreneur. This can be thought of as a control for previous business experience or entrepreneur’s con…dence in their own ability.

I include all individuals that are in the process of the starting a …rm (nascent entrepreneurs) between the ages of 18-65. Although I use individ-ual level data, the survey does not track individindivid-uals over time. One drawback of this is that it is impossible to check for consistency of answers. For ex-ample, if individuals change their mind about the motivation for starting their business. Finally, I use country and time …xed e¤ects to control for any macroeconomic e¤ects which may in‡uence an individual’s motivations for starting a …rm.

It follows that the linear probability model is expressed as M otivationij;t = aj+ t+ Xij;t0 + Incomeij;t+ "ij;t

where M otivationij;t is the dependent binary variable (either opportunity-driven or necessity-opportunity-driven) at time, t; for an individual, i; within a country j. The independent variable of interest is denoted by Incomeij;t, aj denotes country …xed e¤ects, t year …xed e¤ects, and Xij;t0 is a vector of control variables for education, employment, perceived experience and age.

(18)

3.2

Results

Table 2 reports the results of the empirical analysis. The third and fourth columns show the results of the linear probability model when the dependent variable is opportunity-driven and necessity-driven, respectively. Given the binary nature of the dependent variable, it is not surprising that most of the results are inversely related. I report both, however, for convenience purposes.

[Table 2 about here]

The results suggest entrepreneurs are more likely to report their motiva-tion for starting a …rm is opportunity driven when income is larger. Com-pared to poor-income entrepreneurs, having a medium-level income increases the probability of being opportunity driven by 0.049, and having high-level income increases the probability by 0.119. In regards to the other variables, education increases the likelihood of reporting being an opportunity-driven entrepreneur at each level of education. These results are in line with ex-pectations about human capital having a positive e¤ect on innovation and employment opportunities. Being unemployed lowers the probability of be-ing an opportunity-driven entrepreneur by -0.111 compared to a full-time employed individual. This result is unsurprising given that entrepreneur-ship is often used as a necessary route to escape unemployment. Perceived entrepreneurial skill increases the likelihood of an opportunity-driven moti-vation by 0.058. Whether this is interpreted as self-con…dence and/or ac-tual entrepreneurial experience, it is in line with expected positive e¤ects of con…dence and experience needed for opportunity-driven entrepreneurship. Compared to 18-24 year olds, the e¤ect of age decreases the probability of becoming an opportunity-driven entrepreneur, at each age interval. One may …nd these results surprising as older individuals have more experience and have a longer time to amass funds for investment. However, this result could be due to individuals accruing experience in their career as an employee. This would subsequently reduce the incentive to switch to opportunity-driven en-trepreneurship as earnings from paid employment are likely to increase with experience.

There may be a signi…cant di¤erence between nascent entrepreneurs and more established entrepreneurs. Nascent entrepreneurs include all individu-als who engage with the entrepreneurial process. Established entrepreneurs, however, only include entrepreneurs who are successful or who are more will-ing to continue with their project. As a robustness check, I repeat the analysis for individuals who report being owners of an established …rm, as opposed to

(19)

those in the early process of entrepreneurship. The results are given in Table A3 in the Appendix and are similar in both size and e¤ect. The only major di¤erence being whether an individual reports they are unemployed becomes insigni…cant for necessity entrepreneurs. This is to be expected, however, as most owners would categorise themselves as employed when surveyed.

As a …nal robustness check, I split the sample into OECD and non-OECD countries and rerun the analysis for both nascent and established entrepre-neurs. As OECD countries tend to have more generous social institutions compared to non-OECD countries11, it is possible that non-OECD countries are driving the results of the positive e¤ect of income on opportunity-driven entrepreneurship. However, the results given in Table A4 to Table A7 in the Appendix show this not to be the case.

4

Conclusions

This paper has sought to highlight the e¤ects of individual aspirations on the development process. The contribution of this paper is to show how these aspirational e¤ects di¤er depending on the circumstances an individual faces. The model developed within this paper suggests that when individuals share a common aspiration level, to be free from poverty, for example, two di¤erent types of poverty traps occur. Individuals above the poverty line may be su¢ ciently concerned about falling into poverty that they decline investment in wealth-enhancing, risky projects. These individuals invest in safe projects to survive. If individuals are below the poverty line, their aspiration to escape poverty may be so strong that they are willing to invest in risky projects, even if this risky project yields lower returns on average than a less risky alternative.

One conclusion from this model is that those who invest in risky projects, and who do not fear falling into poverty, are opportunity driven. Alterna-tively, those who invest in risky projects who feel they have no other choice, whether to escape poverty, or because they have no other means of income, are necessity driven. As a consequence I test this implication using individ-ual level, cross-county data. The subsequent …ndings suggest the probability of an entrepreneur being opportunity-driven, compared to poor individuals, increases by 0.049 for mid-level income individuals, and 0.119 for high-level income individuals.

One of the strengths of the model presented within this paper is that it can capture many of the results in the existing literature, such as aspiration

11See International Institute of Labour Studies (2014) for a comparison of social

(20)

failure, by introducing a relatively simple form of aspiration level into a standard utility function. However, this model does not discuss the role of e¤ort involving aspirations (as in Dalton, 2016) or how aspirations can evolve endogenously (such as in Genicot and Ray, 2017). Hence, this analysis should be seen as complement to the existing literature surrounding the role of aspirations and poverty traps.

One possible extension of this model is to examine the e¤ects of social insurance on development. In the model described above, a social insur-ance scheme, which prevents individuals falling below the poverty line, could eliminate both poverty traps: individuals, therefore, could no longer be in poverty, or be concerned about falling into poverty. How this social insurance scheme is funded, however, may reduce incentives for individuals to invest in risky, wealth-enhancing projects. This, however, rests on the assumption that our common aspiration level is …xed, such as the aspiration to avoid absolute poverty. If our common aspiration level is to avoid relative poverty, however, this would have implications for the distribution of earnings.

(21)

Appendix

[Table A1 about here] [Table A2 about here] [Table A3 about here] [Table A4 about here] [Table A5 about here] [Table A6 about here] [Table A7 about here]

(22)

References

Ács, Z. J. and A. Varga (2005). "Entrepreneurship, agglomeration and technological change." Small Business Economics 24(3): 323-334.

Antunes, A., T Cavalcanti and A. Villamil (2015). “The E¤ects of Credit Subsidies on Development." Economic Theory 58(1) 1-30.

Åstebro, T., J. Chen and P. Thompson (2011). "Stars and mis…ts: self-employment and labor market frictions." Management Science 57(11): 1999-2017.

Azariadis, C. (1996). "The economics of poverty traps part one: complete markets." Journal of Economic Growth 1(4): 449-486.

Azariadis, C. and J. Stachurski (2005). "Poverty traps." Handbook of Economic Growth 1: 295-384.

Banerjee, A. V. (2000). "The two poverties." Nordic Journal of Political Economy 26(2): 129-141.

Banerjee, A. V. and A. F. Newman (1993). "Occupational choice and the process of development." Journal of Political Economy 101(2): 274-298.

Bernard, T. and A. Seyoum Ta¤esse (2014). "Aspirations: an approach to measurement with validation using ethiopian data." Journal of African Economies 23(2): 189-224.

Blackburn, K. and D. Chivers (2015). "Fearing the worst: the importance of uncertainty for inequality." Economic Theor y 60(2): 345-370.

Bogliacino, F. and I. González-Gallo (2015). "Aspirations, prospect the-ory and entrepreneurship: evidence from Colombia." International Review of Economics 62(3): 271-290.

Carroll, C.D., (2001). "A theory of the consumption function, with and without liquidity constraints." The Journal of Economic Perspectives, 15(3): 23-45.

Dalton, P. S., S. Ghosal and A. Mani (2016). "Poverty and aspirations failure." The Economic Journal 126(590): 165-188.

De Nardi, M., P. Doctor and S.D. Krane (2007). "Evidence on entrepre-neurs in the United States: Data from the 1989-2004" Survey of Consumer Finances.

(23)

Diecidue, E. and J. van de Ven (2008). "Aspiration level, probability of sucess and failure, and expected utility." International Economic Review 49(2): 683-700.

Galor, O. and J. Zeira (1993). "Income distribution and macroeco-nomics." The Review of Economic Studies 60(1): 35-52.

Genicot, G. and D. Ray (2017). "Aspirations and inequality." Economet-rica 85(2): 489-519.

Hamilton, B. H. (2000). "Does entrepreneurship pay? An empirical analysis of the returns to self-employment." Journal of Political Economy 108(3): 604-631.

Herranz, N., S. Krasa and A. Villamil (2015). “Entrepreneurs, Risk Aver-sion and Firm Dynamics.”Journal of Political Economy 123(5), 1133-1176

Hipple, S. (2010). "Self-employment in the United States." Monthly Labor Review 133(9): 17-32.

Holthausen, D. M. (1981). "A risk-return model with risk and return measured as deviations from a target return."The American Economic Re-view 71(1): 182-188.

International Institute of Labour Studies, 2014. "World Social Protection Report 2014/15: Building Economic Recovery, Inclusive Development and Social Justice". International Labour O¢ ce.

Kahneman, D. and A. Tversky (1979). "Prospect theory: An analysis of decision under risk." Econometrica 47(2): 263-291.

Langer, T. and M. Weber (2001). "Prospect theory, mental accounting, and di¤erences in aggregated and segregated evaluation of lottery portfolios." Management Science 47(5): 716-733.

Lin, Z., Picot, G. and Compton, J. (2000). "The entry and exit dynamics of self-employment in Canada." Small Business Economics 15(2): 105-125.

Lopes, L. L. and G. C. Oden (1999). "The role of aspiration level in risky choice: a comparison of cumulative prospect theory and SP/A Theory." Journal of Mathematical Psychology 43(2): 286-313.

Lucas, R. E. (1988). "On the mechanics of economic development." Jour-nal of Monetary Economics 22(1): 3-42.

(24)

Lunati, M., J. M. Schlochtern, and G. Sargsyan (2010). "Measuring entrepreneurship–the OECD-Eurostat Entrepreneurship Indicators Programme." OECD Statistics Brief (15): 1-12.

Macours, K. and R. Vakis (2014). "Changing households’ investment behaviour through social interactions with local leaders: evidence from a randomised transfer programme." The Economic Journal 124(576): 607-633. Maggi, M. A. (2004). "A characterisation of S-shaped utility functions displaying loss aversion." Mimeo Università degli Studi di Pavia.

Mezias, S. J. (1988). "Aspiration level e¤ects: An empirical investiga-tion." Journal of Economic Behavior & Organization 10(4): 389-400.

Mezias, S. J., Y. R. Chen and P. R. Murphy (2002). "Aspiration-level adaptation in an american …nancial services srganization: a …eld study." Management Science 48(10): 1285-1300.

Moav, O. and Z. Neeman (2010). "Status and poverty." Journal of the European Economic Association 8(2-3): 413-420.

Mookherjee, D., S Napel and D. Ray (2010). "Aspirations, segregation, and occupational choice." Journal of the European Economic Association 8(1): 139-168.

Pasquier-Doumer, L. and F. Risso Brandon (2015). "Aspiration failure: a poverty trap for indigenous children in Peru?" World Development 72(C): 208-223.

Payne, J. W. (2005). "It is whether you win or lose: the importance of the overall probabilities of winning or losing in risky choice." Journal of Risk and Uncertainty 30(1): 5-19.

Piketty, T. (1997). "The dynamics of the wealth distribution and the interest rate with credit rationing." The Review of Economic Studies 64(2): 173-189.

Ray, D. (2006). "Aspirations, poverty, and economic change." In: A. V. Banerjee, R. Benabou, and D. Mookherjee, eds. Understanding Poverty. Oxford University Press: 409-421

Ray, D. and A. Robson (2012). "Status, intertemporal choice, and risk-taking." Econometrica 80(4): 1505-1531.

(25)

Romer, P. M. (1986). "Increasing returns and long-run growth." Journal of Political Economy 94(5): 1002-1037.

Schumpeter, J. A. (1934). The theory of economic development.

Wennekers, S., A. van Wennekers, R. Thurik and P Reynolds (2005). "Nascent entrepreneurship and the level of economic development." Small Business Economics 24(3): 293-309.

(26)

Figures and Tables

(27)
(28)
(29)

Table 1: Summary Statistics

Variable Mean S.D.

Nascant Entrepreneur Opportunity Driven 0.541 0.498 Necessity Driven 0.216 0.412 Established Entrepreneur Opportunity Driven 0.478 0.500 Necessity Driven 0.270 0.444

Income Poor Income 0.323 0.468

Midd le Income 0.351 0.477 High Income 0.327 0.469 Education No School 0.035 0.184 Some Secondary 0.281 0.450 Secondary 0.316 0.465 Post Secondary 0.239 0.426 Grad Experience 0.129 0.336 Employment Employed 0.461 0.499 Unemployed 0.073 0.260 Part-time 0.073 0.259 Retired/Disabled 0.065 0.246 Homemaker 0.092 0.288 Student 0.050 0.218

Perceived Experience Entrep. Skill 0.486 0.500

Age 18-24 year olds 0.138 0.345

25-34 year olds 0.213 0.409 35-44 year olds 0.244 0.430 45-54 year olds 0.221 0.415 55-64 year olds 0.183 0.387

(30)

Table 2: Regression results - Nascent Entrepreneurs FE Opportunity Driven Necessity Driven

Income Middle Income 0.049*** -0.047***

(0.012) (0.010)

High Income 0.119*** -0.112***

(0.018) (0.014)

Education Some Secondary 0.006 -0.033

(0.022) (0.023) Secondary 0.043* -0.077*** (0.022) (0.021) Post Secondary 0.075*** -0.126*** (0.023) (0.024) Grad Experience 0.111*** -0.142*** (0.030) (0.027) Employment Unemployed -0.111*** 0.129*** (0.021) (0.019) Part-time -0.011 0.006 (0.011) (0.008) Retired/Disabled -0.040** 0.036** (0.018) (0.016) Homemaker -0.056*** 0.088*** (0.017) (0.015) Student 0.042*** -0.019 (0.014) (0.013)

Percieved Experience Entrep. Skill 0.058*** -0.041***

(0.010) (0.007)

Age 25-34 year olds -0.036*** 0.017**

(0.009) (0.008) 35-44 year olds -0.067*** 0.038*** (0.010) (0.011) 45-54 year olds -0.098*** 0.060*** (0.013) (0.012) 55-64 year olds -0.096*** 0.058*** (0.016) (0.015) Constant 0.404*** 0.379*** (0.026) (0.029)

Year Dummies Yes Yes

County F.E. Yes Yes

Observations 44466 44466

Number of Countries 86 86

Notes: Robust standard errors are reported in parentheses. Signi…cance levels:***p <0.01; **p <0.05; *p <0.1. Low Income, Full Time, No Secondary and 18-24 year olds were dropped from Income, Employment, Education and Age respectively, due to multicollinearity. The independent variable for the 3rd (4th) column is the probability that an entrepreneur reports being opportunity (necessity) driven. The sample only includes nascent entrepreneurs who are de…ned as actively in the process of setting up a business having paid wages <3 months.

(31)

Table A1: Entrepreneurial Data

Data Sorce Hourly worker wages and self-employment earnings

Wage Net Pro…t

Hamilton (2000) U.S., 1984 SIPP Mean 9.63 7.76

S.D. 5.42 10.07

Entrepreneurship rates by education

<High School High School <College College Masters PhD

Hamilton (2000) U.S., 1984 SIPP 12.6 11.1 12.6 15

Hipple (2010) U.S., 2009 CPS 12 11.8 11.8 12.8 13.9

Establishment survival rates SBA (2016) U.S. 2015 BED 80% of establishments started in 2014 survived until 2015

50% of establishments started between 2005 and 2015 last 5 years or longer 33% of establishments started between 2005 and 2015 last 10 years or longer Notes: this table reports data on entrepreneurial earnings, education and survival rates.

Table A2: Sample Countries Sample Countries

Algeria Costa Rica Iran New Zealand* Switzerland*

Angola Croatia Ireland* Nigeria Syria

Argentina Czech Republic* Israel* Norway* Thailand

Australia* Denmark* Italy* Pakistan Tonga

Austria* Dominican Republic Jamaica Panama Trinidad and Tobago

Bangladesh Ecuador Japan* Peru Tunisia

Barbados Egypt Jordan Philippines Turkey

Belgium* El Salvador Kazakhstan Poland* Uganda

Belize Estonia* Latvia* Portugal* United Kingdom*

Bolivia Finland* Lebanon Qatar United States*

Bosnia and Herzegovina France* Libya Russia Uruguay

Botswana Germany* Lithuania Saudi Arabia Venezuela

Brazil Ghana Luxembourg* Senegal Vietnam

Bulgaria Greece* Macedonia Slovakia* Yemen

Burkina Faso Guatemala Malawi Slovenia* Zambia

Cameroon Hong Kong Malaysia South Africa

Canada* Hungary* Mexico* South Korea

Chile* Iceland* Morocco Spain*

China India Namibia Suriname

Colombia Indonesia Netherlands* Sweden*

(32)

Table A3: Regression Results - Established Entrepreneurs FE Opportunity Driven Necessity Driven

Income Middle Income 0.054*** -0.080***

(0.012) (0.011)

High Income 0.134*** -0.156***

(0.015) (0.016)

Education Some Secondary -0.004 -0.031**

(0.013) (0.014) Secondary 0.051*** -0.092*** (0.014) (0.012) Post Secondary 0.081*** -0.141*** (0.017) (0.015) Grad Experience 0.115*** -0.157*** (0.022) (0.018) Employment Unemployed -0.049* 0.033 (0.029) (0.030) Part-time -0.017** -0.008 (0.007) (0.009) Retired/Disabled 0.012 -0.057*** (0.020) (0.019) Homemaker -0.038*** 0.042*** (0.013) (0.015) Student 0.047** -0.102*** (0.022) (0.018)

Percieved Experience Entrep. Skill 0.086*** -0.070***

(0.010) (0.007)

Age 25-34 year olds -0.035*** -0.002

(0.012) (0.010) 35-44 year olds -0.064*** 0.025** (0.013) (0.010) 45-54 year olds -0.096*** 0.053*** (0.014) (0.012) 55-64 year olds -0.111*** 0.056*** (0.014) (0.013) Constant 0.314*** 0.533*** (0.027) (0.020)

Year Dummies Yes Yes

County F.E. Yes Yes

Observations 92830 92830

Number of Countries 86 86

Notes: Robust standard errors are reported in parentheses. Signi…cance levels:***p <0.01; **p <0.05; *p <0.1. Low Income, Full Time, No Secondary and 18-24 year olds were dropped from Income, Employment, Education and Age respectively, due to multicollinearity. The independent variable for the 3rd (4th) column is the probability that an entrepreneur reports being opportunity (necessity) driven. The sample only includes established entrepreneurs who are de…ned as owning and managing a business having paid wages >42 months.

(33)

Table A4: Regression Results - OECD Nascent Entrepreneurs FE Opportunity Driven Necessity Driven

Income Middle Income 0.066*** -0.058***

(0.018) (0.015)

High Income 0.134*** -0.112***

(0.032) (0.024)

Education Some Secondary 0.092* -0.138***

(0.054) (0.049) Secondary 0.100* -0.142*** (0.056) (0.049) Post Secondary 0.126** -0.172*** (0.061) (0.056) Grad Experience 0.154** -0.190*** (0.067) (0.057) Employment Unemployed -0.166*** 0.155*** (0.020) (0.018) Part-time -0.036** 0.001 (0.016) (0.009) Retired/Disabled -0.040** 0.034 (0.019) (0.021) Homemaker -0.100*** 0.121*** (0.025) (0.025) Student 0.032 -0.014 (0.026) (0.017)

Percieved Experience Entrep. Skill 0.056*** -0.039***

(0.018) (0.008)

Age 25-34 year olds -0.043*** 0.021*

(0.014) (0.010) 35-44 year olds -0.082*** 0.052*** (0.020) (0.014) 45-54 year olds -0.115*** 0.069*** (0.026) (0.019) 55-64 year olds -0.119*** 0.065*** (0.027) (0.022) Constant 0.408*** 0.402*** (0.064) (0.058)

Year Dummies Yes Yes

County F.E. Yes Yes

Observations 20443 20443

Number of Countries 29 29

Notes: Robust standard errors are reported in parentheses. Signi…cance levels:***p <0.01; **p <0.05; *p <0.1. Low Income, Full Time, No Secondary and 18-24 year olds were dropped from Income, Employment, Education and Age respectively, due to multicollinearity. The independent variable for the 3rd (4th) column is the probability that an entrepreneur reports being opportunity (necessity) driven. The sample only includes nascent entrepreneurs who are de…ned as actively in the process of setting up a business having paid wages <3 months. Entrepreneurs are only from OECD countries listed in Table A2.

(34)

Table A5: Regression Results - Non-OECD Nascent Entrepreneurs FE Opportunity Driven Necessity Driven

Income Middle Income 0.029* -0.033**

(0.016) (0.014)

High Income 0.096*** -0.104***

(0.022) (0.018)

Education Some Secondary -0.008 -0.008

(0.020) (0.019) Secondary 0.049** -0.074*** (0.023) (0.022) Post Secondary 0.090*** -0.146*** (0.024) (0.025) Grad Experience 0.151*** -0.179*** (0.024) (0.025) Employment Unemployed -0.066*** 0.108*** (0.025) (0.026) Part-time 0.009 0.010 (0.012) (0.012) Retired/Disabled -0.037 0.037 (0.028) (0.023) Homemaker -0.030* 0.071*** (0.017) (0.017) Student 0.052*** -0.024 (0.018) (0.017)

Percieved Experience Entrep. Skill 0.059*** -0.044***

(0.009) (0.010)

Age 25-34 year olds -0.032*** 0.016

(0.012) (0.011) 35-44 year olds -0.056*** 0.027* (0.013) (0.016) 45-54 year olds -0.085*** 0.054*** (0.015) (0.016) 55-64 year olds -0.073*** 0.057*** (0.021) (0.020) Constant 0.364*** 0.406*** (0.025) (0.031)

Year Dummies Yes Yes

County F.E. Yes Yes

Observations 24023 24023

Number of Countries 57 57

Notes: Robust standard errors are reported in parentheses. Signi…cance levels:***p <0.01; **p <0.05; *p <0.1. Low Income, Full Time, No Secondary and 18-24 year olds were dropped from Income, Employment, Education and Age respectively, due to multicollinearity. The independent variable for the 3rd (4th) column is the probability that an entrepreneur reports being opportunity (necessity) driven. The sample only includes nascent entrepreneurs who are de…ned as actively in the process of setting up a business having paid wages <3 months. Entrepreneurs are only from Non-OECD countries listed in Table A2.

(35)

Table A6: Regression Results - OECD Established Entrepreneurs FE Opportunity Driven Necessity Driven

Income Middle Income 0.046*** -0.071***

(0.013) (0.008)

High Income 0.114*** -0.127***

(0.015) (0.013)

Education Some Secondary -0.023 -0.015

(0.019) (0.023) Secondary 0.012 -0.049** (0.022) (0.019) Post Secondary 0.038 -0.084*** (0.024) (0.019) Grad Experience 0.073** -0.099*** (0.031) (0.018) Employment Unemployed -0.085*** 0.073*** (0.019) (0.020) Part-time -0.016 -0.010 (0.011) (0.007) Retired/Disabled -0.004 -0.042** (0.027) (0.017) Homemaker -0.041** 0.023 (0.017) (0.020) Student -0.003 -0.065** (0.038) (0.027)

Percieved Experience Entrep. Skill 0.097*** -0.068***

(0.018) (0.010)

Age 25-34 year olds -0.060** 0.012

(0.024) (0.014) 35-44 year olds -0.095*** 0.035** (0.022) (0.013) 45-54 year olds -0.134*** 0.070*** (0.021) (0.017) 55-64 year olds -0.153*** 0.069*** (0.018) (0.017) Constant 0.420*** 0.396*** (0.049) (0.035)

Year Dummies Yes Yes

County F.E. Yes Yes

Observations 20443 20443

Number of Countries 29 29

Notes: Robust standard errors are reported in parentheses. Signi…cance levels:***p <0.01; **p <0.05; *p <0.1. Low Income, Full Time, No Secondary and 18-24 year olds were dropped from Income, Employment, Education and Age respectively, due to multicollinearity. The independent variable for the 3rd (4th) column is the probability that an entrepreneur reports being opportunity (necessity) driven. The sample only includes established entrepreneurs who are de…ned as owning and managing a business having paid wages >42 months. Entrepreneurs are only from OECD countries listed in Table A2.

(36)

Table A7: Regression Results - Non-OECD Established Entrepreneurs FE Opportunity Driven Necessity Driven

Income Middle Income 0.062*** -0.088***

(0.018) (0.021)

High Income 0.157*** -0.187***

(0.023) (0.025)

Education Some Secondary -0.004 -0.022

(0.014) (0.015) Secondary 0.068*** -0.102*** (0.014) (0.012) Post Secondary 0.110*** -0.181*** (0.018) (0.016) Grad Experience 0.153*** -0.220*** (0.021) (0.022) Employment Unemployed -0.015 0.001 (0.040) (0.039) Part-time -0.013 -0.004 (0.010) (0.016) Retired/Disabled 0.029 -0.072** (0.030) (0.033) Homemaker -0.029 0.045** (0.018) (0.018) Student 0.081*** -0.126*** (0.026) (0.025)

Percieved Experience Entrep. Skill 0.074*** -0.071***

(0.010) (0.010)

Age 25-34 year olds -0.026** -0.008

(0.011) (0.013) 35-44 year olds -0.049*** 0.021 (0.013) (0.013) 45-54 year olds -0.072*** 0.040** (0.016) (0.017) 55-64 year olds -0.080*** 0.052*** (0.020) (0.018) Constant 0.236*** 0.648*** (0.021) (0.028)

Year Dummies Yes Yes

County F.E. Yes Yes

Observations 24023 24023

Number of Countries 57 57

Notes: Robust standard errors are reported in parentheses. Signi…cance levels:***p <0.01; **p <0.05; *p <0.1. Low Income, Full Time, No Secondary and 18-24 year olds were dropped from Income, Employment, Education and Age respectively, due to multicollinearity. The independent variable for the 3rd (4th) column is the probability that an entrepreneur reports being opportunity (necessity) driven. The sample only includes established entrepreneurs who are de…ned as owning and managing a business having paid wages >42 months. Entrepreneurs are only from Non-OECD countries listed in Table A2.

References

Related documents

With 24 percent of total alternative energy filings, carbon capture and storage technologies account for the largest share of UK applications at EPO, followed by hydropower

This study aimed to investigate the associations of serum thyroid stimulating hormone (TSH) levels with cognitive im- pairment and depression after controlling for potential

For ZnO nanofluid, the highest recovery efficiency of 64.26% ROIP with the application of electromagnetic waves and 48.87% ROIP without electromagnetic

In this article, the authors share their experiences about partnership learning and teaching approaches they experienced within the open course ‘Creativity for Learning in Higher

Each pair was also written with the explicit intent to compare and contrast, in the contexts of their references to either race or privilege, the ways in which race enters

expected set of functionalities to cover the complete data life cycle in a personal cloud, namely data collection, storage and recovery, personal cross-computations,

It is interesting to note that although mutual funds and hedge funds seem to disagree on what is the best choice between asset-based and performance-based fees for their

The premium on education is higher in 1998 compared to 1993 at every quantile with one exception - per adult equivalent income is lower in 1998 compared to 1993 at the 10 th